Real Challenge to the architect for architecting and designing for FUTURE application developmen

What do you think?

In my experience...

Now “Software Factory” is the “Buzzword”

To the architect real challenge to make “Platform Independent” auto generated coding tools which will be domain specific.

How extent is this possible? Here things are required, strong domain knowledge along with technical skills. If I talk about .NET initially this came as one of the best programming tools/language/platform – all together the term is used as “Framework”. Yes, thought was very very strong. Vision was Excellent. This is one of the great job done by Microsoft!!! After successfully done this job next it is coming with better approach where basically I found what I used to do in critical way now this is becoming in simpler to the simplest way.

In January 2000 I started programming in PERL, where to write something i.e. to display some thing I used to writer through “print” statement.

In 2000 I started programming in ASP, there I say the way to program is much simpler that the PERL. Where I got the advantages to keep ASP codes were embedded along with the HTML scripts.

But in the same time I used trying to keep the main UI pages in separate file and Codes are into separate files. Same time I used to avoid keeping ASP codes embedded in HTML. That time I used to us FrontPage as page design tool and Interdev as programming tool.

To get maximum benefits I used to follow certain things as follows:

  • Keeping the commonly useable codes as includes files and I used to maintain in the folded naming as “general”. In 2001 I stated leading one team having 10-15 team members. Presently, I remember Jayan, Marina, Libi … and others used to follow the techniques and this is also true that I always open and sometimes they used to guide meJ I coded some program that used to work as template. Sometimes to these developed means my nearer friends used to use the term “Stereotype” work.

  • Next I stated playing with the system tables of SQL Server Databases. From the syscolumns, sysobjects etc. I stated writing program to generated code for Standard Stored Procedures and ASP pages for CRUD operation.

  • Same way I stated working on .NET in the year of 2002 and I have written auto generated code for CRUD operation. Where

    • First I used to select the table and then the columns of it.

    • Then it used to generate the Stored Procedure

    • And from the Stored Procedure to BLL Object (Class) and also UI(.aspx pages)

  • Also I used to use some common code for almost same in so many projects, especially for login, logout, forget password, role management, page accessibilities etc.

After this entire thing what I noticed in .NET 2.0 Microsoft came up Membership API, Personalization Features all these commonly used modules.

Second things related to readabilities of UI and making consistencies there are Master Page, Themes concept are introduced.

Thirdly with concept of DSL now the thinking is coming into new era. Now instead of Membership API, Personalization Features this reusability concept is extending to the Domain Specific Reusable Modules.

I think this is FUTURE. Making reusable modules for the domain specific area, initial coding through the Auto-generated tools, and then customization the whole thing to get the specialized thing.

Finally the coder will start thinking and thinking for code optimization and for the performance issues.

[6951 byte] By [AbuS.Md.Ismail] at [2008-2-4]
# 1
To the architect real challenge to make “Platform Independent” auto generated coding tools which will be domain specific.

I'm not sure that that's true.

Yes, there is definitely a move to using code-generation tools, but these tools so far are limited to relatively simple CRUD operations. DSL initiatives may change this, but that's a way off.

There is an existing 'school' who call themselves the 'Translationists'. They have tools which translate UML diagrams directly into executable code. For more information (actually, the only time I've heard of them) search for postings by H. S. Larman on news://comp.object

I don't know what tools they use, and it seems clear that they operate under some restrictions.

Platform Independance sounds great; as good as Re-usable. I have never needed Platform Independance, though. Re-use was hyped far too much as one of, if not the main, benefits of OO - and by that measure OO has failed. Of course OO hasn't failed; but Re-use remains nearly as hard as ever so in that sense it has.

Good software is very hard to produce. The only silver bullet is to produce as good a design as possible for the implementation in mind bearing in mind things like separation of concerns, focus of design elements (small classes, coherent packages/assemblies, etc), and clarity.

Once DSL's become mainstream, platform independance gained via code-gen from a platform agnostic model becomes easier, I'd imagine. I still see problems: the equivalent of 'experimentors bias' where the agnostic model is designed with the biases (e.g. an OO implementation) of the designer in (unconsious) mind. This would complicate translation to a platform which doesn't support that bias.

But I see that stage being quite a long way off, anyway. And even when we get there I'm not sure how many people will actually require Platform Independance. In 25 years, I've only wished for it twice, both times in the days of minis when the market was far less homologous than it is now and our current platform of choice was going out of business (MAI to DG, then later on with a different system DG to DEC).

simon.s at 2007-9-8 > top of Msdn Tech,Architecture,Architecture General...